Round result of std::fmod(hours, 24) to the nearest float below 24 on double to
float conversion when it is not. Add special type and conversion function along
with tests to be used in all places where such conversion happens.
To avoid producing hours equal to 24 due to double to float precision loss.
conversion from 'const float' to 'int', possible loss of data
conversion from 'double' to 'int', possible loss of data
conversion from 'float' to 'int', possible loss of data
